From:	UUCP%"scgvaxd!cbosgd!okstate!oksma1.!ned" 30-MAY-1985 01:45
To:	ymir!john, engvax!kvc
Subj:	Re: Problems in MAIL

Received: from scgvaxd.UUCP by engvax.UUCP with VMS/UUCP;
          Thu, 30 May 85 01:45:17 PDT
Received: by scgvaxd.UUCP (4.12/4.7)
          id AA10192; Thu, 30 May 85 01:44:59 pdt
Message-Id: <8505300844.AA10192@scgvaxd.UUCP>
Received: from oksma1 by okstate.csnet id a028279; 28 May 85 15:01 CDT
Date: Tue, 28 May 85 14:15 CDT
From: Ned Freed <scgvaxd!ned@oksma1.ARPA>
To: ymir!john, engvax!kvc
Subject: Re: Problems in MAIL

Kevin,
    I think I have fixed your problem with MAIL's use of block mode
I/O. The fix is pretty easy (I guess you're just not an RMS stud).
I managed to get MAIL to produce the appropriate error and checked
out the FAB of the file being sent. It did NOT have the BIO bit set;
it had the BRO bit set instead. In other words, the mode of the file
(block or record) can be changed! However, there are two key restrictions:

(1) All the RAB's must use the same mode.
(2) If a RAB is set to block mode when it is $CONNECT'ed, it will remain
    in block mode until it is disconnected. If the RAB is set to
    record mode when it is $CONNECT'ed, you can pick and choose modes.
    The reason for this is that record mode requires that buffers
    be allocated at $CONNECT time.

Well, MAIL sets the damn RAB to block mode before connecting it, but
while there are sound reasons not to $CLOSE the file, there is no
reason you can't $DISCONNECT the RAB, set the BIO bit in the RAB
properly (there's a BIO bit in both the FAB and the RAB), then reconnect
the RAB to the FAB. The file never opens or closes -- you can play
RAB games pretty much for free.

To make a long story short, just add the following code to MAIL_OUT_CONNECT
just before you set up message_RAB.RAB$L_UBF:

  $DISCONNECT (RAB := message_RAB);
  message_RAB.RAB$L_ROP := uand (message_RAB.RAB$L_ROP, unot (RAB$M_BIO));
  $CONNECT (RAB := message_RAB);

And then MAIL will be a happy little program.

					Ned

P.S. I tried to get the pmdf mailer to have a problem with privilege
     bits being set or reset and confusing MAIL privilege bit checker.
     It never seems to happen when I use PMDF's own enable/disable
     routines, so I guess I'll stick with those.